SERVER-80358 introduced additional synchronization for reopening. Currently it treats archive-based reopening and query-based reopening the same, in that any reopening on a given series conflicts with any other reopening, and any reopening conflicts with any prepared batch on that series. In the case of archive-based reopening, we know exactly which bucket is involved, so we could instead conflict only with prepared batches on the same bucket, other archive-based reopening requests for the same bucket, or query-based reopenings for the series.

This has a potential to resolve regressions for some workloads, but extends the synchronization matrix, so we'll have to evaluate the trade-offs.